CN101178677A - Computer system for protecting software and method for protecting software - Google Patents
Computer system for protecting software and method for protecting software Download PDFInfo
- Publication number
- CN101178677A CN101178677A CNA2007101771355A CN200710177135A CN101178677A CN 101178677 A CN101178677 A CN 101178677A CN A2007101771355 A CNA2007101771355 A CN A2007101771355A CN 200710177135 A CN200710177135 A CN 200710177135A CN 101178677 A CN101178677 A CN 101178677A
- Authority
- CN
- China
- Prior art keywords
- snapshot
- version
- node
- file system
- index node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention provides a creating method of catalogue-grade computer file system snapshot, which comprises: the catalogue that needs to create snap is confirmed; when the catalogue creates the snap firstly, the root index node information of the catalogue work version is added to the snap index list and update the last time point of creating or deleting the snap; a new index node is distributed as the root index node of the creating snap version, and the content of the root index node of the work version is copied to the new distributed index node; the write copy chain list of the work version is copied to the root index node of the new creating snap, and then the write copy chain list content of the work version is deleted; the creating time of snap of the root index node of snap version is set as the present time of the computer system; the root index node of the snap version is linked to the snap chain list of the catalogue. The invention also provides the deleting method of the catalogue-grade computer file system snap and the modifying method of the catalogue work version.
Description
Technical field
The present invention relates to the Snapshot Method of computer file system, particularly the Snapshot Method of computer file system.
Background technology
The computer file system snapping technique has obtained concern more and more widely, and is applied in regulatory requirements such as the backup of file system, release maintenance.Snapshot provides the computer file system consistance impression of certain time point in the past, and by snapshot, the user can realize rolling back to the visit of file system old version and with file system operations such as old version.In addition, as a kind of general means, snapshot can also help the computer file system management software to finish after the backup of online file system, the fault data protection and management functions such as fast quick-recovery effectively.
In local file system, the support of snapshot is divided into file system level and file-level usually.All on the basis of file-level snapshot, provide file system level other snapshot support as these two kinds of file system of ext3cow, elephant, in ZFS, then only supported file system level/subfile system rank snapshot.Under network storage environment, the existing distributed file system then is limited to file system level to the support of snapshot, as VxFS, GPFS, SAN FS etc.Though in SAN FS, also support the snapshot based on subfile system, subfile system can be regarded independently file system as basically, thus remain in essence based on file system level other.
Other snapshot of file system level and the relative merits that the snapshot of file-level all existed separately.Wherein, there is following shortcoming in other snapshot of file system level:
1, coverage is big, when file system level is carried out snapshot associative operation (establishment/deletion/visit), can influence the visit of all users to normal file system, especially under distributed environment, client and visit the coverage that number of processes has been aggravated the snapshot associative operation greatly.
2, lacking dirigibility, is not that all data all have same snapshot demand in the file system, and the user can't create snapshot to the catalogue of needs backup or reservation version according to the actual requirements selectively.
3, consume a large amount of storage resources, behind whole file system establishment snapshot, though adopt COW (Copy on write, Copy On Write) technology can reduce and creates consuming time and avoid the repeated storage of identical data, but all carry out a large amount of consumption that COW still can cause storage resources to the modification of file system.
Then there is following shortcoming in the snapshot of file-level:
1, when file-level is carried out snapshot operation, snapshot adopts log file＜creation-time, deletion time in the version log of index node (inode) and the directory entry usually〉mode constituent act version.Therefore when complying deletion action soon, need the whole file system of traversal could determine which resource to be shared visit, which resource can be recovered, and the efficient of deleting snapshot is very low.
2, compare with other snapshot of file system level, the snapshot of file-level has dirigibility the most widely, but meanwhile, also can bring because the maintenance cost problem that dirigibility caused.Customize different snapshot maintenance strategies as in elephant, permitting a user to each file, but as All Files in the whole file system being carried out the strategy customization, it administers and maintains cost is quite high.For example, the keeper utilizes third party's backup software to carry out online backup work usually, need the backed up data content to leave concentratedly usually under some catalogue, the keeper understands the strategy of snapshot if desired, and according to the backup needs be configured, then keeper's task amount and complexity all can increase greatly.
In sum, there is the excessive defective of snapshot granularity in other snapshot of file system level in the prior art, and there is the defective of snapshot undersized in the snapshot of file-level.
Summary of the invention
Therefore, task of the present invention is to overcome existing Snapshot Method owing to the inappropriate dirigibility that produces of snapshot granule size and the problem on the resource utilization, thereby provides a kind of Snapshot Method that can obtain a better balance on dirigibility during snapshot uses and resource utilization.
To achieve these goals, the invention provides a kind of creation method of computer file system snapshot,, the respective directories behind the establishment snapshot is called the snapshot collection for the assigned catalogue in the computer file system is created snapshot; The establishment of described snapshot may further comprise the steps:
Step 1-1), in computer file system, determine to need create the catalogue of snapshot;
Step 1-2), judge whether described catalogue created snapshot, if described catalogue is for creating for the first time snapshot, then the root index node information of the working version of described catalogue is added in the data entries of snapshot collection concordance list, upgrading the last time point of creating or delete the snapshot collection is the current time of computer system, carry out next step then, otherwise, directly carry out next step;
Step 1-3), distribute the root index node of a new index node as the snapshot version that will create, and with the content replication of the root index node of described working version in newly assigned index node;
Step 1-4, the Copy on write chained list of described working version is copied on the root index node of new establishment snapshot, the Copy on write chained list content with working version empties then;
Step 1-5), the snapshot creation time of the root index node of described snapshot version is set to the current time of computer system;
Step 1-6), the root index node with described snapshot version is linked on the snapshot chained list of described catalogue.
In the technique scheme, described snapshot collection concordance list is used for the index information of all snapshot collection of maintenance calculations machine file system.
In the technique scheme, the described the last time point of creating or delete the snapshot collection is a global information of safeguarding in the whole computer file system, and the time point of described the last establishment or deletion snapshot collection has write down the time that the last snapshot lump quantity changes.
In the technique scheme, include following thresholding in the root index node of described snapshot version: snapshot creation time, nearest Geju City version index, a nearest new version index, by a nearest Geju City version index and a nearest new version index, can be with all versions of corresponding snapshot set by the root index node be linked into a snapshot chained list according to time sequencing.
The present invention also provides a kind of delet method of computer file system snapshot, and the snapshot version for the deletion of the assigned catalogue in computer file system appointment may further comprise the steps:
Step 2-1), in computer file system, in the snapshot chained list of assigned catalogue, select the snapshot version that will delete;
Step 2-2), the adjacent redaction of searching the snapshot version that will delete, inquiry Copy on write chained list on described adjacent redaction obtains the piece that can not be used or the information of index node in adjacent redaction and later release; Described information comprises type, piece number or inode number, date of birth;
Step 2-3), search the adjacent legacy version that to delete the snapshot version, from described adjacent legacy version, find the legacy version snapshot creation time of being write down, if there is not adjacent legacy version in the snapshot version that will delete, the legacy version snapshot creation time that then will search is 0;
Step 2-4), with the date of birth of piece in the Copy on write chained list of described adjacent redaction or index node one by one with step 2-3) Query Result compare, if the date of birth of piece or index node is greater than described Query Result, corresponding piece or the index node resource of the then information of this piece of deletion or index node in the Copy on write chained list, and recovery;
Step 2-5), will wait that the Copy on write chained list of deleting the snapshot version merges in the Copy on write chained list of adjacent redaction;
Step 2-6), will wait to delete the snapshot version and from the snapshot chained list, shift out, and will represent the root index node of adjacent redaction to link to each other with the root index node of the adjacent legacy version of representative.
In the technique scheme, the corresponding snapshot version of described Copy on write chained list, described write copy and writing down pairing snapshot version in the chained list and cause Copy on write because of modification after, the piece that in this version and follow-up version, can not be used or the information of index node, described each piece or each index node information are respectively by tlv triple＜type, piece number or inode number, date of birth〉explain.
The present invention provides a kind of catalogue working version amending method of computer file system snapshot again, may further comprise the steps:
Step 3-1), in computer file system, choose and wait to revise the pairing index node of node;
Step 3-2), according to the timestamp of described index node and the time point of the last establishment or deletion snapshot collection, snapshot collection pointer validity in the described index node is judged,, then carried out next step if described snapshot collection pointer is effective, otherwise, execution in step 3-4);
Step 3-3), the creation-time of the latest snapshot version of more described snapshot collection pointer snapshot set pointed, and the date of birth of waiting to revise node, according to comparative result, whether decision will carry out the Copy on write operation before finishing modification process; If do not need, then directly revise and wait to revise node and finish modification process; Carry out Copy on write operation if desired, after then carrying out Copy on write, revising new node that Copy on write goes out, expand the Copy on write chained list, also need upwards review its father node in the element directory tree at last, re-execute this step;
Step 3-4), upgrade snapshot collection pointer, in described renewal process, by the absolute path name of described node and the snapshot collection information in the snapshot collection concordance list are carried out prefix matching, and upgrade described snapshot collection pointer and timestamp according to matching result.If it is empty upgrading back snapshot collection pointer, then directly revises and wait to revise node and finish modification process; If described snapshot collection pointer is not empty, then re-execute step 3-3).
In the technique scheme, at described step 3-3) in, described according to comparative result, whether decision will carry out the Copy on write operation before finishing modification process comprises:
A), if described date of birth, is not then done Copy on write operation more than or equal to described creation-time, directly revise and wait to revise node and finish modification process;
B), if described date of birth is less than described creation-time, treating the modification node makes the Copy on write operation and revises the new node that described Copy on write goes out, origin node information with the Copy on write correspondence records in the Copy on write chained list of working version then, also need at last the father node of described node in described element directory tree carried out relatively judgement between described date of birth and the described creation-time, if the date of birth of described father node is less than creation-time, then described father node is done the Copy on write operation, and revise the new father node that described Copy on write goes out, the former father node information of Copy on write correspondence is recorded in the Copy on write chained list of working version, repeat said process, after the date of birth of an ancestor node is more than or equal to creation-time, finish modification process.
In the technique scheme, at described step 3-4) in, described result according to coupling, revise snapshot collection pointer and comprise: if coupling is unsuccessful, then described snapshot collection pointer is set to sky; If the match is successful, then snapshot collection pointer is updated to the root inode number of the latest edition in the respective record in the described snapshot collection concordance list.
In the technique scheme, at described step 3-2) in, when the snapshot collection pointer validity in the described index node was judged, if the timestamp of described index node was created more than or equal to described the last time or the time point of deletion snapshot collection, then described snapshot collection pointer was effective.
In the technique scheme, described timestamp is used for judging whether snapshot collection state changes.
In the technique scheme, described element directory tree is that the 4 class fundamental elements that comprise index node, a location piece, directory block, data block in the computer file system are carried out obtaining after abstract, in abstraction process, with the non-leaf node of described index node, a location piece, directory block and so on abstract be a kind of abstract node, described abstract node comprises this nodal information and child node descriptor.
In the technique scheme, described child node descriptor comprises the child node date of birth, and child node is described and the child node pointer.
The invention has the advantages that:
1, the present invention is by providing a kind of method of supporting computer file system directory level snapshot; can effectively satisfy user and application program in actual applications; to catalogue and the file that needs carry out data protection, online backup or release maintenance, carry out the demand of operations such as snapshot creation, visit, deletion neatly.Compare the snapshot support of file system level, the present invention can solve snapshot operation to influences such as the time of whole file system access, the file that relates to and catalogue scopes effectively by the snapshot granularity of dwindling, and significantly reduces a large amount of consumption problems of storage resources.Compare the snapshot support of file-level, the present invention can reduce the management maintenance cost that file-level snapshot granularity is brought effectively by the snapshot granularity that enlarges, and can solve the problem of deletion efficiency technically.
2, a kind of computer file system method of directory level snapshot of supporting provided by the invention specifically comprises establishment snapshot, deletion snapshot and revises the catalogue working version.Create in the Snapshot Method, only relate to the read-write operation of the individual data structure in the internal memory, this makes that creating snapshot operation almost finishes in moment.In the deletion Snapshot Method, avoided the process of the index node chained list traversal that the deletion of file-level snapshot brings, and operation also is data structure in the internal memory, so deletion action is also very fast.Revise in the catalogue working version method, the shared tree that adopts the COW technology effectively to postpone between the version is revised the process that divides because of generation, has not only saved the time that copy is spent, the consumption of more having saved storage resources.
Description of drawings
Below, describe embodiments of the invention in conjunction with the accompanying drawings in detail, wherein:
Fig. 1 is a file system directories level snapshot institutional framework synoptic diagram of the present invention;
Fig. 2 is the abstract schematic of non-leaf node object;
Fig. 3 is a rough schematic of element directory tree;
Fig. 4 is the state change map in the snapshot creation process;
Fig. 5 is the state change map in the snapshot delete procedure;
Fig. 6 is the process flow diagram of the creation method of computer file system snapshot of the present invention;
Fig. 7 is the process flow diagram of the delet method of computer file system snapshot of the present invention;
Fig. 8 is the process flow diagram of the catalogue working version amending method of computer file system snapshot of the present invention.
Embodiment
The present invention is described further below in conjunction with the drawings and specific embodiments.
The present invention has realized the support to snapshot in order to overcome the excessive or too small defective of prior art snapshot granularity on the directory level of file system.Compare with the file-level snapshot with the file system level snapshot, the directory level snapshot has different characteristics, at first the involved related notion of directory level snapshot is described below.
1, snapshot collection (snapset): in computer file system, the user can distinguish with the catalogue of not creating snapshot for the catalogue that will create snapshot mutually to the corresponding snapshot of directory creating, and the catalogue that the present invention will create snapshot is called the snapshot collection.Comprise all sub-directories and son file in the catalogue in the content that snapshot set comprised.For example, created snapshot for a catalogue "/root/usr/dir ", this catalogue is exactly a snapshot collection so, all sub-directories that comprised in this catalogue as "/root/usr/dir/file1 ;/root/usr/dir/sub_dir2/file2 " etc., all belong to this snapshot collection as "/root/usr/dir/sub_dir1 ;/root/usr/dir/sub_dir2/grand_dir3 " etc., son file.
2, LMTS: a global variable of in whole computer file system, safeguarding, the last time point of creating or deleting the snapshot collection of this global variable record (Last Modify Time ofSnapset, LMTS).
3, version, working version and snapshot version: any catalogue in the file system all has version.For the catalogue of never creating snapshot, a working version is only arranged, described working version correspondence the traditional file system access application entity that does not possess snapshot functions.And each snapshot of a snapshot set is all distinguished corresponding snapshot version.Do not possess dependence between the version in logic, for example: a snapshot collection has a working version and a snapshot version, and after working version was deleted, the snapshot version still existed.
4, snapshot collection concordance list: snapshot collection concordance list is used for the index information of all snapshot collection of maintaining file system.Any one snapshot collection all possesses corresponding indexed information items in concordance list:＜directory name, the root inode number of latest edition 〉.Usually latest edition is a working version, for example, in accompanying drawing 1,, can locate the root index node of corresponding snapshot collection "/root/usr/dir " working version in " file system tree " by "/root/usr/dir " in the snapshot collection concordance list 1 corresponding index information.By another "/root/home/a/b " corresponding index information in the table, then can locate the root index node of snapshot collection "/root/home/a/b " working version.When the working version of snapshot collection did not exist, such as deleted, latest edition just referred to a up-to-date snapshot version so.Snapshot collection concordance list is the distinguishing characteristics of the present invention and file system level Snapshot Method or file-level Snapshot Method, after having adopted snapshot collection concordance list, the snapshot of can having known in the computer file system which directory creating, which catalogue is not then created snapshot, and can realize creating searching fast of snapshot catalogue.
5, the root index node of version: the root index node of any version all has following special thresholding, snapshot creation time (epoch), nearest Geju City version index (prev), a nearest new version index (next), above-mentioned special thresholding marks in accompanying drawing 4.
By the pointers such as prev, next in the version root index node, can be with all versions of corresponding snapshot set by the root index node be linked into a snapshot chained list according to time sequencing.When only having a version, corresponding prev, next pointer are sky.In accompanying drawing 1, the different editions that catalogue is generated in creating snapshot processes illustrates.Snapshot collection "/root/usr/dir " has three versions: working version, snapshot version B and snapshot version A.By the respective pointer in three version root index nodes, with the index node formation snapshot chained list of three versions.
And, from the root index node of any version, name resolving just can travel through the content of this version by basic file system, promptly it descends all son files, sub-directory etc.Each version of a snapshot set can form one tree independently.For example, shown in 2 in the accompanying drawing 1, the working version of/root/usr/dir/ has just formed one tree, and this tree is the stalk tree of whole file system simultaneously.Owing to adopt COW (copy-on-write, Copy on write) technology raising access efficiency and reduction capacity to take, share branch so exist usually between the tree of different editions.For example, from the root index node of working version travel through downwards/working version of root/usr/dir set 2 o'clock, the new branch 21 of addressable COW, and working version respectively and snapshot version B share branch 22; Equally, sharing branch 3 is shared branches of snapshot version B and two trees of snapshot version A.
6, Copy on write chained list (COW_list): each version all has a COW_list, cause COW because of modification when writing down this version in this chained list as working version, thus the resource (piece and index node) that in this version and follow-up version, can not be used.Every above-mentioned resource, all corresponding tlv triple in COW_list:＜type, piece number/inode number, date of birth 〉, wherein, type is meant that this resource is a piece or for index node, and piece number/inode number then is the unique identification of this resource in file system, date of birth is meant the time point that this resource is assigned with and uses, by and file system create constantly between at interval a second number identify.By the date of birth in the tlv triple, and COW_list the version creation-time (epoch) in the renewal root index node of corresponding version just can determine the life cycle of resource (piece and index node): this resource is at＜date of birth, version creation-time〉time period in be used.
The COW_list of the general catalogue working version of non-snapshot collection is empty, because COW did not take place.To a directory creating snapshot time, the COW_list of catalogue working version will be transferred on the snapshot version of new establishment, and the COW_list of working directory is emptied.In catalogue is to create under the situation of snapshot for the first time, so be sky because the COW_list of COW working version never takes place before, so after the COW_list switching, the COW_list of the new snapshot version of creating also just is empty, that is to say that the COW_list one of the snapshot version that any one catalogue obtains is decided to be sky after creating snapshot for the first time.In addition, only the tlv triple number that writes down among the COW_list of working version just might become many because the COW process takes place, so the COW process can not take place because the snapshot version is affirming of can not being modified.
In accompanying drawing 5, form and content to COW_list have been done relevant explanation: writing down 4 resources that can not be used among the COW_list of " the adjacent redaction " shown in 61 in this version and later release thereof, as one＜B wherein, 20,120〉the expression piece number is 20, and date of birth is 120 piece; Writing down 3 among the COW_list of " snapshot to be deleted " shown in 62 can be by the resource that is used in this version and the later release (as " adjacent redaction ") thereof, as one＜I wherein, 90,60〉the expression inode number is 90, and date of birth is 60 index node.
7, file system element directory tree structure: in computer file system, all nodes can be divided into 4 class fundamental elements according to type, are respectively: index node, a location piece (indirect block), directory block (dentry block), data block.In the snapshot document system, the direct control entity that above-mentioned 4 dvielements also are COW in the snapshot document system.Above-mentioned 4 class fundamental elements are done further abstractly in the snapshot document system, can organize becomes file system element directory tree.
In file system element directory tree, node can be divided into non-leaf node and leaf node by type.Non-leaf node comprises index node, a location piece, directory block three classes.Leaf node generally is data block (Data block), but is that 0 o'clock index node or first directory block in the empty list also can become leaf node in file size.The non-leaf node of all 3 classes can abstractly be a kind of Object node, and it has two category informations: this nodal information and child node descriptor.This nodal information comprises the nodal community relevant information; Child node is described and comprised: the child node date of birth, child node is described and the child node pointer.In accompanying drawing 2, non-leaf node is described.Wherein, comprise two category informations relevant in the nodal community relevant information of index node with snapshot.The first kind is aforesaid snapshot creation time (epoch), nearest Geju City version index (prev), a nearest new version index (next) in the index node.For the root index node of snapshot collection version, these three items are effective, and to common index node, these three values are invalid.Second category information comprises a snapshot collection pointer (snapset ptr) and a timestamp (timestamp), under the effective situation of snapshot collection pointer, point to the root index node of the latest edition of the affiliated snapshot collection of this index node, if its value is empty, represent that then this index node does not belong to a snapshot collection.Timestamp then is a necessary condition judging snapshot collection pointer validity, judges by comparing timestamp and LMTS whether snapshot collection pointer is effective, and then snapshot collection pointer is effective greater than LMTS for timestamp, otherwise invalid.
Describe and pointer by the child node in the above-mentioned all kinds of non-leaf node, certain version tissue of certain catalogue of file system can be become an element directory tree, and can organize whole file system by these element directory trees.By the element directory tree, when handling the snapshot associative operation, can ignore the type of concrete non-leaf node in logic, and carry out unified description and processing.In the flow process of " to the modification of catalogue working version " for example mentioned below, there is the judgement of " trace back and judge whether father node needs COW " among the step 3-2 (two), wherein " father node " must be a non-leaf node, is the non-leaf node of which kind of type but need not to pay close attention to.Fig. 3 is the synoptic diagram of a simple element directory tree, in the figure, from the index node of "/root/usr/dir ", it is non-leaf node of each layer and leaf node in the inventory element tree of root (can be the root of any one version of this catalogue) that the child node pointer by each node layer can travel through with "/root/usr/dir " step by step.
After the feature that the directory level fast photographic system is had is done above-mentioned explanation, describe respectively to the establishment of directory level snapshot, deletion and to the modification of catalogue working version.
With reference to figure 6, and, the visioning procedure of directory level snapshot is done following explanation in conjunction with example:
Step 1-1, according to user's selection, need to create the catalogue of snapshot in the decision file system;
Whether step 1-2, judgement will be created the catalogue of snapshot and whether create snapshot, be the snapshot collection promptly; If create snapshot for the first time, then in the snapshot collection concordance list of computer file system, add the directory entry relevant with the catalogue that will create snapshot, upgrading LMTS is the current time of computer system, carry out next step then, if not creating snapshot for the first time, then directly carry out next step;
Step 1-3, new index node conduct of distribution are newly created the root index node of snapshot version, and the root index node content of working version are copied to the root index node of new establishment snapshot;
Step 1-4, the COW_list of working version is copied on the root index node of new establishment snapshot, and the COW_list content of working version is emptied;
Step 1-5, the snapshot creation time (epoch) that will newly create the root index node of snapshot version are set to the current time of computer system;
Step 1-6, the root index node that will newly create the snapshot version are linked on the snapshot chained list of corresponding catalogue.Newly assigned index node is when link, if current directory is to create snapshot for the first time, the root index node that then has only the catalogue working version in the snapshot chained list of catalogue, the root index node that newly assigned index node directly is linked to working version gets final product, in link process, will newly distribute the forwarding pointer of root index node point to the root index node of working version, back with the root index node of working version to newly assigned index node of pointed; If created snapshot before the current directory, a then newly assigned index node is when link, the forwarding pointer of newly assigned index node is pointed to the root index node of working version, back with the root index node of working version to newly assigned index node of pointed, and back to the adjacent old edition origin of pointed index node with newly assigned index node, the forwarding pointer of adjacent old edition origin index node points to newly assigned index node.
In accompanying drawing 4, under the condition of having created snapshot (this catalogue has been the snapshot collection) before the catalogue, the process that the root index node of new establishment snapshot version is linked describes in detail.Before to the new snapshot of some directory creatings, the working version that a snapshot version A and this catalogue self have been arranged, pairing inode number of snapshot version A is 80, pairing inode number of working version curr is 10, the root index node of the legacy version index point snapshot version A correspondence in the root index node of working version, the new version index of the root index node of snapshot version A is pointed to the root index node of working version.After the redaction to the directory creating snapshot, the redaction of the snapshot of creating is snapshot version B: at first the root inode number of distributing for snapshot version B be 100 and with the content replication of index node 10 to wherein; Then the Copy on write chained list shown in 41 being transferred to becomes 52 and the Copy on write chained list of working version emptied on the snapshot version B, shown in 51; Then, the creation-time (epoch) that snapshot version B is set is the current time 150 of system; At last, the new version index that snapshot version B is set is pointed to the root index node of working version, the root index node of its legacy version index point snapshot version A is set, in the root index node of snapshot version A and working version, carry out corresponding modification, thereby realize linking between the snapshot chained list of the root index node of new establishment snapshot and catalogue.In this figure, all contents that need increase newly or revise are indicated by the twill frame.
The user can realize the deletion to snapshot according to actual needs after creating snapshot, with reference to figure 7, and in conjunction with example, the deletion flow process of snapshot is elaborated:
Step 2-1, in the snapshot chained list of a certain catalogue the selected snapshot version that will delete;
Step 2-2, search the adjacent redaction that to delete the snapshot version, from adjacent redaction, inquire about COW_list.In this step, described adjacent redaction is meant and compares renewal with the snapshot version that will delete and the working version time interval is the shortest snapshot version or catalogue.If described catalogue only exists the snapshot version that will delete or the snapshot version that will delete except working version be up-to-date snapshot version, then adjacent redaction is the working version of catalogue, if described catalogue is except working version, the snapshot version that also has other, and have the snapshot version of comparing renewal with the snapshot version that will delete in these versions, then adjacent redaction is the more recent version the most approaching in time with the snapshot version that will delete.
In practical operation, searching the adjacent redaction that will delete the snapshot version only needs to search forwarding pointer root index node pointed on the root index node of the snapshot version correspondence that will delete, and the version of this root index node representative is exactly described adjacent redaction.In aforementioned explanation to COW_list, the content that COW_list comprised is explained, can know " piece " or " index node " that is not used under COW_list version and the later release from COW_list.These " pieces " or " index node " that are recorded among the COW_list are also referred to as " piece " or " index node " that " is killed " by this version.
Step 2-3, search the adjacent legacy version that will delete the snapshot version, and inquire about the snapshot creation time (epoch) that adjacent legacy version writes down, if there is not adjacent legacy version, then Query Result is 0.When the adjacent legacy version of inquiry, can find pairing index node of adjacent legacy version to pointer according to deleting the back of snapshot version root index node.The snapshot creation time that adjacent legacy version write down is put down in writing accordingly at the snapshot creation time of its root index node (epoch) Xiang Zhongyou.If the back is NULL to pointer, prove that then adjacent legacy version does not exist, creation-time (epoch) is then thought the time point that file system is created so, promptly 0.
Step 2-4, the date of birth of piece among the COW_list of adjacent redaction and index node is compared with step 2-3 Query Result item by item, if described date of birth is greater than described Query Result, then delete the corresponding tlv triple among the COW_list, and reclaim corresponding piece or index node resource, make these resources available again;
Step 2-5, will wait that the COW_list that deletes the snapshot version merges among the COW_list of adjacent redaction;
Step 2-6, will wait to delete the snapshot version and from the snapshot chained list, shift out, and will represent the root index node of adjacent redaction to link to each other with the root index node of the adjacent legacy version of representative.
Accompanying drawing 5 is an instantiation of snapshot deletion, and in this example, a snapshot chained list has three snapshot versions, and the snapshot version that delete is " snapshot to be deleted " among the figure.Have three by its piece that " kills " among the COW_list of " adjacent redaction " among the figure, be＜B 20,120 respectively,＜B, 21,70 〉,＜B, 22,80〉and an index node＜I who " is killed " by adjacent redaction, 100,110 〉.The snapshot creation time of " adjacent legacy version " is 100 among the figure.At first, draw by the date of birth of every resource among the COW_list and the snapshot time ratio of " adjacent legacy version ": two resources that meet the deletion condition are only arranged in the COW_list of " adjacent redaction ", be respectively＜B, 20,120〉and＜I, 100,110 〉, so the resource of deletion above-mentioned two and release block numbers 20 and inode number 100 correspondences.Then, three resource record items in will the COW_list of " snapshot to be deleted " all merge among the COW_list of adjacent redaction, corresponding root index node of deletion " snapshot to be deleted " in the snapshot chained list at last, and will remain two versions and be linked.
Except above-mentioned snapshot creation and snapshot deletion flow process, the present invention also comprises the modification to the catalogue working version.The modification of catalogue working version is meant any content in the tree of the working version correspondence of revising this catalogue, such as, in the accompanying drawing 12 be exactly/ one tree that the working version of root/usr/dir/ forms, and revise any content in the tree and all be called as modification the catalogue working version.
In modification process, also may cause the operation of COW to the catalogue working version.For the ease of understanding, at first the basic thought to the COW operation is illustrated.
Catalogue a has two child node b and c, and a is a root node.Suppose a is created snapshot, duplicate a and become a ' in constructive process, a ' is the root node of the snapshot version of new establishment.This moment, a pointed to own child node b and c, because a ' duplicates, so a ' also points to child node b and c, promptly b and c are the resources that is shared.During the visit working version, promptly can have access to all the elements of working version by the pointer of a, a sensing child node b, the pointer of a sensing child node c.In like manner, during visit snapshot version, pass through all the elements that pointer that a ', a ' point to child node b, pointer that a ' points to child node c also can have access to the snapshot version.
When working version need be revised, from 1 to 10 totally 10 values were for example arranged among the b originally, need will be wherein 2 be revised as 20, but can not directly revise b, because b is also being quoted by snapshot a ', visit at any time b node content among the snapshot a ' must be all the same (content remains 1 to 10), if revise b, the modification of working version is successful so, and the content of snapshot version is just destroyed.So, need carry out a COW operation this moment: at first duplicate a b ' (promptly new b ' of distribution, the content of b is copied among the b ', 10 values of from 1 to 10 are arranged) among the b ', the content of b ' is made amendment (as 2 among the b ' is revised as 20) then, claim that b is by the source node of COW, the new node that b ' goes out for COW.Behind COW, the child node pointer modified of again a being pointed to b is for pointing to b '.Through aforesaid operations, c still shares, but b and b ' have just separated, and b ' are made amendment afterwards again, does not just need the COW operation, because the b ' of this moment only belongs to working version.
After the thought to the COW operation describes, below the flow process of revising the catalogue working version is carried out specific description.May relate to the modification of catalogue working version that a plurality of nodes all need to make amendment in the inventory element tree, and following only the description revised the wherein flow process of certain node, because this node can be non-leaf node of any class or leaf node.
All there are a snapshot collection pointer (snapset ptr) and a timestamp (timestamp) in all index nodes, in accompanying drawing 2, be present in the file attribute of index node, under the effective situation of this snapset ptr, point to the root index node of the latest edition of the affiliated snapshot collection of this index node, if its value is empty, represent that then this index node does not belong to a snapshot collection.Timestamp then is a necessary condition judging snapset ptr validity, by relatively timestamp and LMTS judge whether snapset ptr is effective, LMTS represents time point newly-increased recently or deletion snapshot collection, then snapset ptr is effective greater than LMTS for timestamp, otherwise invalid.
With reference to figure 8, the modification process of catalogue working version is as follows:
3-1, obtain and wait to revise the pairing index node of node, if node to be revised is exactly an index node, the index node that then obtains is exactly himself;
3-2, if the timestamp in the index node greater than LMTS, and snapset ptr be empty, then do not need COW, directly revises node to be revised, the modification process end;
In this step, " timestamp is greater than LMTS " provable snapset ptr is effective, and under effective situation, snapset ptr is that sky illustrates that then waiting to revise node does not belong to a snapshot collection, problem with regard to not existing snapshot version and working version to share does not also just need to carry out the COW operation so.That is: when any catalogue that does not belong to the snapshot collection is made amendment, all do not need the COW operation.
3-3, if index node in timestamp greater than LMTS, and snapset ptr is not empty, illustrate that node to be revised belongs to the snapshot collection that this snapset ptr points to, then relatively the creation-time (epoch) and both sizes of date of birth (being recorded in its father node) of waiting to revise node of the up-to-date snapshot version of this snapshot set:
(1) date of birth does not then need COW more than or equal to epoch, directly revises node to be revised, and modification process finishes.
In this case, illustrate that waiting to revise node had carried out the COW operation, only belong to working version, can directly make amendment, need not to do again the COW operation.For example: the new node b ' that COW goes out in the preamble COW operational instances, its date of birth certainly greater than the epoch of latest snapshot at that time, do not need COW when then revising b ', directly modification getting final product.
(2) date of birth then needs COW and revises node to be revised less than epoch; And need trace back and judge whether father node needs COW, if the date of birth of its father node is less than epoch, then father node also needs COW, carrying out also need passing up to after the modification of father node grandfather's node carries out COW and judges and revise, up to the date of birth of an ancestor node more than or equal to epoch, just stop to trace back, modification process finishes.
In this case, illustrate that node to be revised shared by working version and up-to-date snapshot version, therefore need carry out COW.And needing to trace back father node in this step is because parent directory may also be shared, if share, it points to the pointer and the date of birth of sub-directory to parent directory owing to need modification so, so also just need COW.In like manner need to pass up to the ancestor node crossed of COW before this always.This traces back the root index node that process might pass up to this working version.
In addition, any node that carries out COW operation, all need with by the origin node of COW＜type, node number, date of birth〉record among the COW_list of catalogue working version.
3-4, if the timestamp in the index node less than LMTS, the newly-increased of snapshot collection or deletion took place in expression recently, promptly snapset ptr may be invalid, so need upgrade snapset ptr.Renewal process is by carrying out prefix matching with the absolute path name of this node and the snapshot collection information in the snapshot collection concordance list:
(1) if do not mate, prove that then at this moment waiting to revise node does not belong to any snapshot collection, need the timestamp and the snapset ptr that upgrade index node to be set to sky, do not carry out COW and operate, direct modification node to be revised, modification process finishes;
(2) if can mate, then proof node to be revised this moment is present in a snapshot set, so what content the snapset ptr no matter is, all it is updated to the root inode number of the latest edition in the respective record in the concordance list, and the timestamp of renewal index node, then skip to step 3-3.
In this step, the modification of the timestamp of index node is different from the common modification to index node, need carry out the judgement of COW whether and father node and process such as trace back.
It should be noted last that above embodiment is only unrestricted in order to technical scheme of the present invention to be described.Although the present invention is had been described in detail with reference to embodiment, those of ordinary skill in the art is to be understood that, technical scheme of the present invention is made amendment or is equal to replacement, do not break away from the spirit and scope of technical solution of the present invention, it all should be encompassed in the middle of the claim scope of the present invention.
Claims (13)
1. the creation method of a computer file system snapshot for the assigned catalogue in the computer file system is created snapshot, is called the snapshot collection with the respective directories behind the establishment snapshot; The establishment of described snapshot may further comprise the steps:
Step 1-1), in computer file system, determine to need create the catalogue of snapshot;
Step 1-2), judge whether described catalogue created snapshot, if described catalogue is for creating for the first time snapshot, then the root index node information of the working version of described catalogue is added in the data entries of snapshot collection concordance list, upgrading the last time point of creating or delete the snapshot collection is the current time of computer system, carry out next step then, if described catalogue is not to create snapshot for the first time, directly carry out next step;
Step 1-3), distribute the root index node of a new index node as the snapshot version that will create, and with the content replication of the root index node of described working version in newly assigned index node;
Step 1-4), the Copy on write chained list of described working version is copied on the root index node of new establishment snapshot, the Copy on write chained list content with working version empties then;
Step 1-5), the snapshot creation time of the root index node of described snapshot version is set to the current time of computer system;
Step 1-6), the root index node with described snapshot version is linked on the snapshot chained list of described catalogue.
2. the creation method of computer file system snapshot according to claim 1 is characterized in that, described snapshot collection concordance list is used for the index information of all snapshot collection of maintenance calculations machine file system.
3. the creation method of computer file system snapshot according to claim 1, it is characterized in that, the described the last time point of creating or delete the snapshot collection is a global information of safeguarding in the whole computer file system, and the time point of described the last establishment or deletion snapshot collection has write down the time that the last snapshot lump quantity changes.
4. the creation method of computer file system snapshot according to claim 1, it is characterized in that, include with lower threshold value in the root index node of described snapshot version: snapshot creation time, nearest Geju City version index, a nearest new version index, by a nearest Geju City version index and a nearest new version index, can be with all versions of corresponding snapshot set by the root index node be linked into a snapshot chained list according to time sequencing.
5. the delet method of a computer file system snapshot is deleted the snapshot version of appointment for the assigned catalogue in the computer file system, may further comprise the steps:
Step 2-1), in computer file system, in the snapshot chained list of assigned catalogue, select the snapshot version that will delete;
Step 2-2), the adjacent redaction of searching the snapshot version that will delete, inquiry Copy on write chained list on described adjacent redaction obtains the piece that can not be used or the information of index node in adjacent redaction and later release; Described information comprises type, piece number or inode number, date of birth;
Step 2-3), search the adjacent legacy version that to delete the snapshot version, from described adjacent legacy version, find the legacy version snapshot creation time of being write down, if there is not adjacent legacy version in the snapshot version that will delete, the legacy version snapshot creation time that then will search is 0;
Step 2-4), with the date of birth of piece in the Copy on write chained list of described adjacent redaction or index node one by one with step 2-3) Query Result compare, if the date of birth of piece or index node is greater than described Query Result, corresponding piece or the index node resource of the then information of this piece of deletion or index node in the Copy on write chained list, and recovery;
Step 2-5), will wait that the Copy on write chained list of deleting the snapshot version merges in the Copy on write chained list of adjacent redaction;
Step 2-6), will wait to delete the snapshot version and from the snapshot chained list, shift out, and will represent the root index node of adjacent redaction to link to each other with the root index node of the adjacent legacy version of representative.
6. the delet method of computer file system snapshot according to claim 5, it is characterized in that, the corresponding snapshot version of described Copy on write chained list, described write copy and writing down pairing snapshot version in the chained list and cause Copy on write because of modification after, the piece that in this version and follow-up version, can not be used or the information of index node, described each piece or each index node information is respectively by tlv triple＜type, piece number/inode number, date of birth〉explain.
7. the catalogue working version amending method of a computer file system snapshot may further comprise the steps:
Step 3-1), in computer file system, choose and wait to revise the pairing index node of node;
Step 3-2), according to the timestamp of described index node and the time point of the last establishment or deletion snapshot collection, snapshot collection pointer validity in the described index node is judged,, then carried out next step if described snapshot collection pointer is effective, otherwise, execution in step 3-4);
Step 3-3), the creation-time of the latest snapshot version of more described snapshot collection pointer snapshot set pointed, and the date of birth of waiting to revise node, according to comparative result, whether decision will carry out the Copy on write operation before finishing modification process; If do not need, then directly revise and wait to revise node and finish modification process; Carry out Copy on write operation if desired, after then carrying out Copy on write, revising new node that Copy on write goes out, expand the Copy on write chained list, also need upwards review its father node in the element directory tree at last, re-execute this step;
Step 3-4), upgrade snapshot collection pointer, in described renewal process, by the absolute path name of described node and the snapshot collection information in the snapshot collection concordance list are carried out prefix matching, and upgrade described snapshot collection pointer and timestamp according to matching result.If it is empty upgrading back snapshot collection pointer, then finish renewal process; If described snapshot collection pointer is not empty, then re-execute step 3-3).
8. the catalogue working version amending method of computer file system snapshot according to claim 7 is characterized in that, at described step 3-3) in, described according to comparative result, whether decision will carry out the Copy on write operation before finishing modification process comprises:
A), if described date of birth, is not then done Copy on write operation more than or equal to described creation-time, directly revise and wait to revise node and finish modification process;
B), if described date of birth is less than described creation-time, treating the modification node makes the Copy on write operation and revises described node, origin node information with the Copy on write correspondence records in the Copy on write chained list of working version then, also need at last the father node of described node in described element directory tree carried out relatively judgement between described date of birth and the described creation-time, if the date of birth of described father node is less than creation-time, then described father node is done the Copy on write operation, and modification father node, the former father node information of Copy on write correspondence is recorded in the Copy on write chained list of working version, repeat said process, after the date of birth of an ancestor node is more than or equal to creation-time, finish modification process.
9. the catalogue working version amending method of computer file system snapshot according to claim 7, it is characterized in that, at described step 3-4) in, described result according to coupling, revising snapshot collection pointer comprises: if coupling is unsuccessful, then described snapshot collection pointer is set to sky; If the match is successful, then snapshot collection pointer is updated to the root inode number of the latest edition in the respective record in the described snapshot collection concordance list.
10. according to the catalogue working version amending method of claim 7 or 8 or 9 described computer file system snapshots, it is characterized in that, at described step 3-2) in, when the snapshot collection pointer validity in the described index node is judged, if the timestamp of described index node is more than or equal to the described the last time point of creating or deleting the snapshot collection, then described snapshot collection pointer is effective.
11. the catalogue working version amending method of computer file system snapshot according to claim 7 is characterized in that, described timestamp is used for judging whether snapshot collection state changes.
12. the catalogue working version amending method of computer file system snapshot according to claim 7, it is characterized in that, described element directory tree is that the 4 class fundamental elements that comprise index node, a location piece, directory block, data block in the computer file system are carried out obtaining after abstract, in abstraction process, with the non-leaf node of described index node, a location piece, directory block and so on abstract be a kind of abstract node, described abstract node comprises this nodal information and child node descriptor.
13. the catalogue working version amending method of computer file system snapshot according to claim 12 is characterized in that described child node descriptor comprises the child node date of birth, child node is described and the child node pointer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007101771355A CN100541441C (en) | 2007-11-09 | 2007-11-09 | A kind of snapshot creation of computer file system, deletion and amending method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2007101771355A CN100541441C (en) | 2007-11-09 | 2007-11-09 | A kind of snapshot creation of computer file system, deletion and amending method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101178677A true CN101178677A (en) | 2008-05-14 |
| CN100541441C CN100541441C (en) | 2009-09-16 |
Family
ID=39404938
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2007101771355A Active CN100541441C (en) | 2007-11-09 | 2007-11-09 | A kind of snapshot creation of computer file system, deletion and amending method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100541441C (en) |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101299197B (en) * | 2008-06-30 | 2011-09-28 | 无锡中星微电子有限公司 | Method and unit for adding and deleting task |
| CN102339321A (en) * | 2011-11-09 | 2012-02-01 | 上海盛霄云计算技术有限公司 | Network file system with version control and method using same |
| CN102799607A (en) * | 2012-05-31 | 2012-11-28 | 新奥特（北京）视频技术有限公司 | Method for processing data in snapshot mode |
| CN103038767A (en) * | 2010-07-30 | 2013-04-10 | 国际商业机器公司 | Data sharing for file clones |
| CN103336828A (en) * | 2013-07-05 | 2013-10-02 | 深圳市康拓普信息技术有限公司 | Reading and writing-in method for real-time database |
| CN103473250A (en) * | 2012-06-05 | 2013-12-25 | 国际商业机器公司 | Method and system for preserving past states of file system nodes |
| CN103473277A (en) * | 2013-08-27 | 2013-12-25 | 华为技术有限公司 | Snapshot method and device for file systems |
| CN103473251A (en) * | 2012-06-05 | 2013-12-25 | 国际商业机器公司 | File system snapshots method and system using selective tuple versioning in acomputing environment using processing devices |
| CN103678040A (en) * | 2013-12-06 | 2014-03-26 | 北京明朝万达科技有限公司 | Snapshooting and backspacing method and system based on computer system |
| CN104714755A (en) * | 2013-12-13 | 2015-06-17 | 华为技术有限公司 | Snapshoot management method and device |
| CN104933133A (en) * | 2015-06-12 | 2015-09-23 | 中国科学院计算技术研究所 | Metadata snap storage and access method in distributed file system |
| CN105095016A (en) * | 2014-05-16 | 2015-11-25 | 北京云巢动脉科技有限公司 | Disk snapshot roll-back method and device |
| CN105095015A (en) * | 2014-05-13 | 2015-11-25 | 北京云巢动脉科技有限公司 | Disk snapshot establishment method and apparatus |
| CN105843905A (en) * | 2010-11-22 | 2016-08-10 | 日立数据系统工程英国有限公司 | Device and method for managing file system object of file system |
| CN106326040A (en) * | 2016-08-27 | 2017-01-11 | 浪潮(北京)电子信息产业有限公司 | Method and device for managing snapshot metadata |
| CN106547926A (en) * | 2016-12-14 | 2017-03-29 | 北京华云网际科技有限公司 | The optimization method and device of the root node of the snapshot tree of distributed block storage |
| CN107180092A (en) * | 2017-05-15 | 2017-09-19 | 中国科学院上海微系统与信息技术研究所 | A kind of control method of file system, device and terminal |
| CN107277111A (en) * | 2017-05-19 | 2017-10-20 | 腾讯科技（深圳）有限公司 | A kind of company-data processing method, master node, slave node and cluster |
| CN107526840A (en) * | 2017-09-14 | 2017-12-29 | 郑州云海信息技术有限公司 | File system snapshot querying method, device and computer-readable recording medium |
| CN108829813A (en) * | 2018-06-06 | 2018-11-16 | 郑州云海信息技术有限公司 | A kind of File Snapshot method and system based on distributed memory system |
| CN109308231A (en) * | 2017-07-28 | 2019-02-05 | 中国石油化工股份有限公司 | A kind of management method of spatial data version |
| CN109358947A (en) * | 2018-10-08 | 2019-02-19 | 郑州云海信息技术有限公司 | A kind of realization method and system of virtual machine snapshot |
| CN109445989A (en) * | 2018-10-22 | 2019-03-08 | 郑州云海信息技术有限公司 | A kind of file system data backup method and device |
| WO2019091085A1 (en) * | 2017-11-13 | 2019-05-16 | 华为技术有限公司 | Snapshot comparison method and apparatus |
| CN110515543A (en) * | 2019-08-02 | 2019-11-29 | 星辰天合（北京）数据科技有限公司 | Object bucket-based snapshot method, device and system |
| CN111158860A (en) * | 2019-12-30 | 2020-05-15 | 咪咕文化科技有限公司 | Data operation method, electronic device and storage medium |
| CN111538864A (en) * | 2020-03-25 | 2020-08-14 | 新华三技术有限公司合肥分公司 | Method and device for reducing Buildrun consumption |
| CN112698956A (en) * | 2021-01-19 | 2021-04-23 | 腾讯科技（深圳）有限公司 | Data object processing method, device, equipment and storage medium |
| CN112860487A (en) * | 2021-02-09 | 2021-05-28 | 山东英信计算机技术有限公司 | Method, system, device and medium for copying volume shadow |
| CN113535642A (en) * | 2021-08-05 | 2021-10-22 | 统信软件技术有限公司 | File searching method and computing device |
| CN115292248A (en) * | 2022-09-30 | 2022-11-04 | 中孚安全技术有限公司 | Data cleaning method, system and equipment based on multiple data versions |
| CN112698956B (en) * | 2021-01-19 | 2024-04-12 | 腾讯科技（深圳）有限公司 | Data object processing method, device, equipment and storage medium |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103268270B (en) * | 2013-05-10 | 2016-04-13 | 曙光信息产业(北京)有限公司 | The management method of snapshot and device |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7072916B1 (en) * | 2000-08-18 | 2006-07-04 | Network Appliance, Inc. | Instant snapshot |
| US20030093443A1 (en) * | 2001-11-15 | 2003-05-15 | Huxoll Vernon F. | System and method for creating online snapshots |
| CN1266609C (en) * | 2003-06-05 | 2006-07-26 | 联想(北京)有限公司 | Fine granule quick camera carry-out method |
| US7440966B2 (en) * | 2004-02-12 | 2008-10-21 | International Business Machines Corporation | Method and apparatus for file system snapshot persistence |
| CN100405310C (en) * | 2005-07-12 | 2008-07-23 | 中国科学院计算技术研究所 | Fast photographic system and method |
-
2007
- 2007-11-09 CN CNB2007101771355A patent/CN100541441C/en active Active
Cited By (50)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101299197B (en) * | 2008-06-30 | 2011-09-28 | 无锡中星微电子有限公司 | Method and unit for adding and deleting task |
| CN103038767A (en) * | 2010-07-30 | 2013-04-10 | 国际商业机器公司 | Data sharing for file clones |
| CN103038767B (en) * | 2010-07-30 | 2016-01-27 | 国际商业机器公司 | For the method and system of the data sharing of file clone |
| CN105843905B (en) * | 2010-11-22 | 2019-06-11 | 日立数据管理有限公司 | For managing the device and method of the file system object in file system |
| CN105843905A (en) * | 2010-11-22 | 2016-08-10 | 日立数据系统工程英国有限公司 | Device and method for managing file system object of file system |
| CN102339321A (en) * | 2011-11-09 | 2012-02-01 | 上海盛霄云计算技术有限公司 | Network file system with version control and method using same |
| CN102799607A (en) * | 2012-05-31 | 2012-11-28 | 新奥特（北京）视频技术有限公司 | Method for processing data in snapshot mode |
| CN103473251A (en) * | 2012-06-05 | 2013-12-25 | 国际商业机器公司 | File system snapshots method and system using selective tuple versioning in acomputing environment using processing devices |
| CN103473250A (en) * | 2012-06-05 | 2013-12-25 | 国际商业机器公司 | Method and system for preserving past states of file system nodes |
| US9747317B2 (en) | 2012-06-05 | 2017-08-29 | International Business Machines Corporation | Preserving past states of file system nodes |
| CN103473251B (en) * | 2012-06-05 | 2017-03-01 | 国际商业机器公司 | The method and system of the file system snapshot using selectivity tuple versioned is provided in the computing environment using processing apparatus |
| US9569458B2 (en) | 2012-06-05 | 2017-02-14 | International Business Machines Corporation | Preserving a state using snapshots with selective tuple versioning |
| CN103473250B (en) * | 2012-06-05 | 2016-08-17 | 国际商业机器公司 | For preserving the method and system of the past state of file system nodes |
| CN103336828A (en) * | 2013-07-05 | 2013-10-02 | 深圳市康拓普信息技术有限公司 | Reading and writing-in method for real-time database |
| CN103473277B (en) * | 2013-08-27 | 2017-04-05 | 华为技术有限公司 | The Snapshot Method and device of file system |
| CN103473277A (en) * | 2013-08-27 | 2013-12-25 | 华为技术有限公司 | Snapshot method and device for file systems |
| CN103678040A (en) * | 2013-12-06 | 2014-03-26 | 北京明朝万达科技有限公司 | Snapshooting and backspacing method and system based on computer system |
| CN104714755A (en) * | 2013-12-13 | 2015-06-17 | 华为技术有限公司 | Snapshoot management method and device |
| CN104714755B (en) * | 2013-12-13 | 2020-01-03 | 华为技术有限公司 | Snapshot management method and device |
| CN105095015A (en) * | 2014-05-13 | 2015-11-25 | 北京云巢动脉科技有限公司 | Disk snapshot establishment method and apparatus |
| CN105095015B (en) * | 2014-05-13 | 2018-02-23 | 北京云巢动脉科技有限公司 | The method for building up and device of a kind of disk snapshot |
| CN105095016A (en) * | 2014-05-16 | 2015-11-25 | 北京云巢动脉科技有限公司 | Disk snapshot roll-back method and device |
| CN105095016B (en) * | 2014-05-16 | 2018-05-18 | 北京云巢动脉科技有限公司 | A kind of disk snapshot rollback method and device |
| CN104933133A (en) * | 2015-06-12 | 2015-09-23 | 中国科学院计算技术研究所 | Metadata snap storage and access method in distributed file system |
| CN104933133B (en) * | 2015-06-12 | 2018-09-07 | 中国科学院计算技术研究所 | Meta-data snap in distributed file system stores and accesses method |
| CN106326040A (en) * | 2016-08-27 | 2017-01-11 | 浪潮(北京)电子信息产业有限公司 | Method and device for managing snapshot metadata |
| CN106547926A (en) * | 2016-12-14 | 2017-03-29 | 北京华云网际科技有限公司 | The optimization method and device of the root node of the snapshot tree of distributed block storage |
| CN107180092A (en) * | 2017-05-15 | 2017-09-19 | 中国科学院上海微系统与信息技术研究所 | A kind of control method of file system, device and terminal |
| CN107180092B (en) * | 2017-05-15 | 2020-10-23 | 中国科学院上海微系统与信息技术研究所 | File system control method and device and terminal |
| CN107277111A (en) * | 2017-05-19 | 2017-10-20 | 腾讯科技（深圳）有限公司 | A kind of company-data processing method, master node, slave node and cluster |
| CN107277111B (en) * | 2017-05-19 | 2018-09-04 | 腾讯科技（深圳）有限公司 | A kind of company-data processing method, master node, slave node and cluster |
| CN109308231A (en) * | 2017-07-28 | 2019-02-05 | 中国石油化工股份有限公司 | A kind of management method of spatial data version |
| CN107526840A (en) * | 2017-09-14 | 2017-12-29 | 郑州云海信息技术有限公司 | File system snapshot querying method, device and computer-readable recording medium |
| CN110018989A (en) * | 2017-11-13 | 2019-07-16 | 华为技术有限公司 | A kind of method and apparatus that snapshot compares |
| WO2019091085A1 (en) * | 2017-11-13 | 2019-05-16 | 华为技术有限公司 | Snapshot comparison method and apparatus |
| CN110018989B (en) * | 2017-11-13 | 2021-05-18 | 华为技术有限公司 | Snapshot comparison method and device |
| CN108829813A (en) * | 2018-06-06 | 2018-11-16 | 郑州云海信息技术有限公司 | A kind of File Snapshot method and system based on distributed memory system |
| CN109358947A (en) * | 2018-10-08 | 2019-02-19 | 郑州云海信息技术有限公司 | A kind of realization method and system of virtual machine snapshot |
| CN109445989A (en) * | 2018-10-22 | 2019-03-08 | 郑州云海信息技术有限公司 | A kind of file system data backup method and device |
| CN110515543A (en) * | 2019-08-02 | 2019-11-29 | 星辰天合（北京）数据科技有限公司 | Object bucket-based snapshot method, device and system |
| CN110515543B (en) * | 2019-08-02 | 2021-02-19 | 星辰天合（北京）数据科技有限公司 | Object bucket-based snapshot method, device and system |
| CN111158860B (en) * | 2019-12-30 | 2024-02-23 | 咪咕文化科技有限公司 | Data operation method, electronic device and storage medium |
| CN111158860A (en) * | 2019-12-30 | 2020-05-15 | 咪咕文化科技有限公司 | Data operation method, electronic device and storage medium |
| CN111538864A (en) * | 2020-03-25 | 2020-08-14 | 新华三技术有限公司合肥分公司 | Method and device for reducing Buildrun consumption |
| CN112698956A (en) * | 2021-01-19 | 2021-04-23 | 腾讯科技（深圳）有限公司 | Data object processing method, device, equipment and storage medium |
| CN112698956B (en) * | 2021-01-19 | 2024-04-12 | 腾讯科技（深圳）有限公司 | Data object processing method, device, equipment and storage medium |
| CN112860487A (en) * | 2021-02-09 | 2021-05-28 | 山东英信计算机技术有限公司 | Method, system, device and medium for copying volume shadow |
| CN113535642A (en) * | 2021-08-05 | 2021-10-22 | 统信软件技术有限公司 | File searching method and computing device |
| CN115292248A (en) * | 2022-09-30 | 2022-11-04 | 中孚安全技术有限公司 | Data cleaning method, system and equipment based on multiple data versions |
| CN115292248B (en) * | 2022-09-30 | 2023-01-03 | 中孚安全技术有限公司 | Data cleaning method, system and equipment based on multiple data versions |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100541441C (en) | 2009-09-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100541441C (en) | A kind of snapshot creation of computer file system, deletion and amending method | |
| US11704290B2 (en) | Methods, devices and systems for maintaining consistency of metadata and data across data centers | |
| CA2533916C (en) | File system represented inside a database | |
| US10254996B1 (en) | Fast migration of metadata | |
| US5519855A (en) | Summary catalogs | |
| US7979441B2 (en) | Method of creating hierarchical indices for a distributed object system | |
| US5812130A (en) | Data management system and method for concurrent engineering | |
| US8255373B2 (en) | Atomic multiple modification of data in a distributed storage system | |
| US10262004B2 (en) | Native snapshots in distributed file systems | |
| CN100498781C (en) | Method for storing metadata of logic document system by adhesion property | |
| CN102663076B (en) | Method for processing file data | |
| US20060248128A1 (en) | Efficient mechanism for tracking data changes in a database system | |
| US7620668B2 (en) | Authoritative and non-authoritative restore | |
| CN110209528B (en) | Data backup method, device, server and storage medium | |
| CN101866305A (en) | Continuous data protection method and system supporting data inquiry and quick recovery | |
| US20210064264A1 (en) | Efficient restoration of content | |
| CN102693174B (en) | Object broker database constraint-oriented backup method | |
| ZA200510092B (en) | File system represented inside a database | |
| Mohan | A database perspective on lotus domino/notes | |
| Kozlova et al. | Efficient consistency support for distributed mobile applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |